Introduction

Cross-electrophile coupling is a significant complement to the classic cross-coupling reactions, in which two structurally diverse and bench-stable electrophiles can be coupled directly in the presence of an external reductant1,2,3,4,5,6,7. This strategy bypasses the preparation and handling of organometallic reagents with limited stability and commercial availability and is a sought-after technology for synthetic practitioners. In general, C−C bond formation in the cross-electrophile coupling reactions commonly occurs at a specific activated site (e.g., Cl, Br, I, etc.) preinstalled on the two coupling partners8,9,10,11,12,13,14. Accordingly, a complementary method capable of new C(sp2)–C(sp3) bond construction at remote, unfunctionalized sites in readily available feedstocks, which will allow straightforward access to scaffolds that would otherwise be difficult to prepare, is still in high demand for synthetic chemists. More recently, migratory cross-coupling enabled by “NiH-chain-walking” with 1,n-hydrogen-atom shift has emerged as the most studied method to address this challenging issue (Fig. 1a)15,16,17,18,19. Mechanistically, a two-electron β-hydride elimination/migratory insertion occurs, thus accomplishing the migration of nickel species from a preactivated to a cross-coupling site (finally to a benzylic site)20,21,22. Despite significant progress in this regard, stereocontrolled migratory coupling between two halide-based electrophiles remains a challenge23.

Fig. 1: Reaction design for asymmetric migratory cross-electrophile coupling by radical π-system-independent 1,2-N-shift.
figure 1

a Typical migratory cross-coupling enabled by “NiH-chain walking” (previous studies). b Typical 1,n-Imino radical shift (Frey-1990, Han-2021). c Our design: Migratory cross-coupling with Ni/CPA cooperative catalysis. d Previous studies to access optimally pure bioactive β-amino acid motifs (For example). e Enantioselective migratory cross-electrophile coupling via 1,2-N-shit (This work).

Heteroatom migration/rearrangement is one of the most potent technologies for innovating new reactions and increasing molecular complexity in organic and pharmaceutical synthesis24,25,26,27,28. Over the past decade, new synthetic methods that comprise radial heteroatom migration involving the 1,n-shift of B-29,30,31,32, Si-33,34,35, P-36, and halo groups37, have been intensively exploited, delivering rapid access to heteroatom molecular scaffolds that could be in high demand by synthetic chemists. Nitrogen-based functionalities play vital roles as bioelements in bioorganic and organic chemistry. For example, the migration of radical N-functionalities occurs naturally in the lysine 1,2-aminomutase-catalyzed interconversion of L-α-lysine into L-β-lysine in Costridia38. Although N-rearrangement reactions also show high utilization prospects for organic and pharmaceutical chemistry39,40,41, radical-mediated migration of N-centered functionalities remains underdeveloped. To the best of our knowledge, migratory cross-electrophile coupling enabled by a formal 1,2-N-shift has also not been reported to date. Several challenging issues must be addressed in this scenario: (a) The methods are limited to a specific radical precursor that furnishes a nitrogen-based π-system (e.g., imine, azide)42,43,44,45, wherein a mechanism pattern that comprises a radical ipso-addition to an imino/azide π-bond to form an unstable, strained aziridinyl radical is commonly involved, thereby enabling a β-fragmentation toward formal 1,2-amino migration (Fig. 1b). Therefore, abundant and readily available organic amines without N-π-systems commonly display less reactivity in these documented N-migration reactions; (b) To our knowledge, however, migratory cross-electrophile coupling by a π-system-independent 1,n-N-shift, between two halide-based electrophiles, has never been exploited. Admittedly, several challenges remain to be addressed. First, the radical 1,2-N-shift process is readily disturbed by neophyl-type rearrangement46, hydrogen atom transfer (HAT) or single-electron-transfer (SET) oxidation39, instead of funnelling into a subsequent cross-coupling catalytic cycle. Secondly, the competition of intramolecular C-H cyclization47, direct cross-coupling and homocoupling at the preactivated sites prior to amino migration exist commonly (i.e., lacking regioselectivity); (c) Thirdly, nickel-catalyzed C(sp2)-C(sp3) cross-coupling via radical heteroatom migration has always led to stereoablation48,49, let alone this version of migratory coupling onto an α-position of an ester group which easily gives rise to racemization under basic conditions because of the newly formed stereogenic center50,51,52. Therefore, enantioselective control of migratory coupling (especial at a α-position of an ester group) via radical 1,2-N-shift remains a challenge. Encouraged by our recent progress in radical nitrogen migration cascades, we speculated that the nascent C(sp3)-centered radical arising from the 1,2-N-shift of a β-bromo α-amino acid ester, by the judicious design of Ni/L catalytic systems, could funnel into a subsequent cross-coupling catalytic cycle with aryl halides. Furthermore, in view of these significant advances in asymmetric metal catalysis combined with organocatalysis53,54,55,56, dual catalysis combining a Ni(II)/chiral ligand complex and chiral Brønsted acid could dictate the enantioselectivity of this migratory C(sp2)-C(sp3) cross-coupling and produce diverse enantioenriched β-amino acid derivatives (Fig. 1c).

Enantioenriched non-natural β-amino acids and analogues serve as important building blocks for bioactive scaffolds, which exist in many medicinal molecules, such as Taxol, pan-Akt inhibitors, and Jasplakinolide57,58. Therefore, the development of an efficient and enantioselective synthetic technique for β-amino acid derivatives has attracted much interest from organic and medicinal chemists59,60,61. Despite previous studies, there remains much to explore. Firstly, radical migration strategies enabling the catalytic enantio- and regioselective synthesis of β-amino acid derivatives remains underexplored62,63,64; furthermore, multiple synthetic steps and problematic chiral separation by preparative high-performance liquid chromatography (HLPC) is commonly required for this target (Fig. 1d)65. Secondly, although skeletal editing of abundant and easily accessible α-amino acids via a radical 1,2-N-shift sequence represents a promising approach to enantioenriched α-arylated β-amino acid esters, the method has never been successful. Given our ongoing interest in migratory coupling66,67, we aimed to highlight the instance of migratory cross-electrophile coupling via radical 1,2-amino migration, which can be used to access β-amino acid motifs. Moreover, upon cooperative catalysis of the Ni(II) catalyst/ chiral bisoxazoline (Box) complex and chiral phosphoric acid (CPA), this tandem radical sequence accomplished enantioselective control of C(sp3)-C(sp2) cross-coupling (Fig. 1e). Inspired by Gong’s pioneering work68, the readily available and safe diboron reagent was employed as the reducing agent in our reductive coupling system. Taken together, this strategy exhibits high enantio-/diastereo-/regio-selectivity and extensive substrate compatibility, and operates avoiding use of stoichiometric quantities of unsafe metal reductants (e.g., Zn, Mn powders), toxic radical initiators, and/or elevated temperatures, thus allowing for a late-stage synthesis of non-natural β-amino acids and analogues.

Results

Reaction development

We initially investigated the optimal conditions for the cascade 1,2-nitrogen migration/cross-coupling reaction using the serine-derived bromide (1) and 4-bromobenzonitrile (2) as model racemic substrates (Table 1). After extensive investigation of the reaction parameters (see the ‘Optimization of reaction conditions’ section in Supplementary information for details), we discovered that the N-migratory coupling product (3) could be obtained in 74% yield when NiBr2•dme (10 mol%), 2-(4,5-dihydro-1H-imidazol-2-yl)−4-methylpyridine L1 (12 mol%), and B2neo2 (2 equiv.) were used as catalyst, ligand, and reductant, respectively. By contrast, almost no regioselective isomer (3′), that can be formed via direct coupling of the two pre-halogenated sites, was detected by gas chromatography (GC)–MS or 1H NMR. Moreover, the relative configuration of the as-obtained product (3) was unambiguously confirmed by X-ray crystallography (CCDC 2383157 (for 3), CCDC 2383157 (for 6) and CCDC 2383157 (for 23) contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service). The steric effect of the aromatic rings, which occur as a paired group on the N atom, contributes to suppressing the alternative intramolecular C-H cyclization toward tetrahydroisoquinoline-4-carboxylates. Accordingly, serine-derived bromides with a pair of 2,6-disubstituted benzyl groups on the N atom were designed as electrophilic coupling partners. Notably, organic diboron served as the reductant appeared to be crucial for promoting the 1,2-N-shift because using stoichiometric Zn or Mn as the reductant led to 1,4-aryl migration to afford the Smiles-type rearrangement isomer 3′ (entries 1–3). In addition, B2neo2 displayed the highest efficiency among the diboron reductants, including B2pin2 and B2(cat)2 (entries 4–5)68,69,70. Other pyridine-imidazoline ligands, such as ligands L2–L3, were also evaluated, the results showed that 2-(4,5-dihydro-1H-imidazol-2-yl)−4-methylpyridine L1 remained the optimal choice (entries 6–7). Notably, bipyridine ligands L4L6 exhibited unparallel efficiency in the coupling reactions (entries 8–10).

Table 1 Optimization of reaction conditionsa

Next, we evaluated the feasibility of enantioselective migratory cross-coupling. Among the collection of chiral N-based ligands (mainly Box and Bilm ligands) for this enantioselective migratory coupling with rac-1, the cyclo-bridge Box framework proved to be the most promising (see Table S2 in the Supplementary Information for details). Furthermore, a five-membered cyclo-Box ligand bearing an aryl substituent exhibited much higher reactivity and enantioselectivity (entries 11–15). Further evaluation of the substituents on the cyclo-bridge Box ligands revealed that both the electronic properties and steric hindrance had a significant impact on the reactivity and enantioselectivity. For example, the m-CF3 substituted Box ligand led to an inferior yield and enantioselectivity, whereas the 2,6-dimethyl substituted cyclo-Box ligand (L10) furnished the desired product 3 in 64% yield with 85:15 e.r. (entries 12 and 14). Thus encouraged, we further optimized the catalytic systems by attempting a cooperative catalysis between the Ni/Box ligand complex and a chiral Brønsted acid. Gratifyingly, suitable cooperation between CPA and the chiral Box ligand in the catalytic systems could contribute to the formation of a sterically defined chiral pocket71, which furnished 3 with higher enantioselectivity (entries 16–19). Intriguingly, cooperation between the (S,S)-Box ligand (L10) and (R)-CPA 2 proved to be optimal for the enantiomeric ratio; however, an alternative combination of the (S,S)-Box ligand and (S)-CPA hampered the enantioselectivity in this migratory coupling (entry 18). Such a stereocontrolled divergence of different combinations of Box Ligand and CPA stereoisomers for the rearrangement reactions appears to be consistent with Gong’s observation of cooperative catalysis by the Pd/chiral phosphoramidite ligand complex and CPA72. Notably, CPA could, theoretically, easily be transformed into its sodium salts under basic conditions (using Na2CO3), which would really function as the as a labile ligand55,56, according to its asymmetric catalytic efficiency. In addition, replacing NiBr2•dme with NiCl2•dme resulted in a slight decrease in the reactivity and enantioselectivity of this model reaction (entry 20).

Substrate scope

With the optimal reaction conditions established, the scope of aryl bromides used in this migratory cross-coupling reaction was examined under racemic conditions (Fig. 2). Gratifyingly, the migratory coupling reactions of serine-derived bromide 1a with an array of activated aryl bromides, including 4-bromobenzonitrile, methyl 4-bromobenzoate, and 1-(4-bromophenyl)-ethenone, proceeded smoothly, leading to the formation of structurally diverse β-amino acid motifs in moderate to good yields. In addition, this set of reactions exhibited excellent regioselectivity toward the N-migratory coupling products (products 36). The configuration of product 3 and 6 was unambiguously confirmed by X-ray crystallography (CCDC 2383157 (for 3), CCDC 2383157 (for 6) and CCDC 2383157 (for 23) contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service). Notably, sterically hindered 2-bromobenzonitrile and methyl 2-bromobenzoate were shown to be viable coupling partners in this coupling reaction (products 8 and 9). In addition, unactivated aryl bromides such as aniline and phenolic ester derivatives were well tolerated in this migratory reaction (products 1722). Migratory coupling also appears to be compatible with structurally diverse heteroaryl bromides and thus provides successful results for the introduction of N-heterocyclic motifs such as pyridine, quinoline, benzofuran, thiazole and etc., which are in high demand in medicinal chemistry (products 2331). Furthermore, the amino acid ethyl/benzyl ester was a viable precursor for the migratory coupling reaction (products 3233). Next, we evaluated the reactivity of these radical precursors with different benzyl groups tethered to the N atom and observed that diverse pairs of benzyl groups, such as o-F/p-ClC6H4CH2, 2,6-Cl2C6H4CH2, 2,6-F2C6H4CH2, and others, were well tolerated under the optimal conditions (products 3440), albeit in moderate yield. Notably, intramolecular C-H cyclization occurs simultaneously for substrates bearing less sterically hindered benzyl groups (products 34–35). In addition, we investigated the performance of α-amino acid-derived secondary bromides as the electrophiles in the coupling reactions. An array of secondary bromides was compatible with the catalytic conditions and led to structurally diverse β-amino acid esters and analogues (products 41–44) in moderate yield with excellent diastereoselectivity (usually d.r. > 20:1). Furthermore, when aryl bromides derived from several bioactive molecule such as helicid, were examined, all the desired product 45 were obtained in moderate yields and with good regioselectivity. To further demonstrate the synthetic utility of this method, another type of radical coupling precursor (amino alcohol-derived tosylates) was examined as feedstock under standard conditions. As expected, the migratory cross-coupling performed well and delivered the pharmaceutically important β-arylethylamines 46. In addition, considering the ready availability of inexpensive serine derivatives, we investigated the compatibility of serine tosylates instead of serine bromides under these reaction conditions. Delightedly, we discovered that these serine tosylates or mesylate with aryl bromides underwent sequential rearrangement and cross-coupling to afford the desired products in moderate yields. Of note, the N-protecting benzyl group can be removed by Pd(OH)2/C/H2 catalytic systems. Nevertheless, serine tosylate/iodine with the tosyl or CBz group on N-atom instead of a benzyl-type protecting group was less effective for migratory coupling. In addition, we further investigated the 1,2-O-migration reaction by employing a structurally analogous O-migratory precursor as the substrate; however, this approach did not yield the desired product.

Fig. 2: Scope of (hetero)aryl bromides and N-benzyl amino acid esters in Ni-catalyzed migratory couplinga.
figure 2

aReaction conditions in “Method A: alkyl bromide 1 (0.2 mmol), (hetero)aryl bromide 2 (2 equiv), NiBr2•dme (10 mol%), L1 (12 mol%), B2neo2 (2 equiv), H2O (2 equiv) K2CO3 (2 equiv) and CH3CN/EtOH (v: v = 1.8: 0.2, 0.1 M) at 50 °C under argon atmosphere for 30 h; isolated yields. b By-products arising from intramolecular C-H cyclization. c Diastereoselective ratio (dr) was determined by 1H NMR and/or 19 F NMR analysis.

Subsequently, we performed a preliminary evaluation of the efficiency of the newly established asymmetric catalytic systems of the Ni/chiral Box complex and CPA for the stereocontrol of migratory cross-electrophile coupling (Fig. 3). The serine-derived primary bromide 1 underwent migratory cross-coupling with structurally diverse aryl bromides in moderate yields with excellent enantioselectivity, thus resulting in stereocontrolled construction of a C(sp3)-C(sp2) bond toward enantioenriched α-arylated β-amino acids (products 3, 4, 5, 6, 47, 20, 32, and 48). Besides, the unactivated aryl bromides was well tolerated under the asymmetric catalytic conditions (products 17, 18, 21, and 49). Notably, the formation of the alkyl-alkyl dimerization products, as an isomeric mixture of the homo/cross-dimerization of radicals Int-III and Int-1 possibly as shown in Fig. 4, was observed to form by HRMS, as well as the homo-dimerization of aryl bromide, these existing side reactions would impress the yield of desired products to some extent, which appears to be responsible for the moderate yield outcome in many cases (usually <70% yields). Heteroaryl bromides have also proved to be compatible with asymmetric catalytic conditions, thus incorporating pharmaceutically interesting heterocyclic moieties into these amino acid skeletons, albeit with sightly decreasing enantioselectivity (products 25 and 50). Also, a pair of benzyl group 2,6-F2C6H4CH2 on the N-atom, instead of 2,6-Me2C6H4CH2, were observed to be viable precursor for migratory coupling, thus accessing enantioenriched β-amino acid esters bearing pharmaceutically important fluorine atom (products 37, 38, 39, 40, and 51). In addition, the natural molecular moieties were successfully and stereospecifically installed on these amino acid scaffolds using this migratory coupling strategy (dr >20:1; product 45). Undoubtedly, the abovementioned successes would endow this asymmetric catalysis strategy with broad utilization prospects for the synthesis of complex non-natural amino acids and analogues24,25,26,27,28. Furthermore, by using this migratory cross-coupling strategy, a pan-Atk inhibitor analogue can be successfully prepared in two steps with excellent enantioselectivity, avoiding multiple stepwise syntheses (especially challenging chiral separation steps), as previously reported65,73.

Fig. 3: Ni/CPA-catalyzed enantioselective migratory cross-electrophile coupling by 1,2-N-shifta.
figure 3

aReaction conditions in “Method B”: alkyl bromide 1 (0.1 mmol), (hetero)aryl bromide 2 (2 equiv.), NiBr2•dme (10 mol%), L10* (12 mol%), (R)-CPA-2 (10 mol%), B2neo2 (2 equiv), H2O (2 equiv) Na2CO3 (3 equiv) and CH3CN (1 mL, 0.1 M) at 40 °C under argon atmosphere for 36 h; Enantiomeric ratios (e.r.) were determined using chiral HLPC; isolated yields. b (11bR)−4-Hydroxy-2,6-bis[4-(2-naphthalenyl)phenyl]−4-oxide dinaphtho[2,1-d:1’,2’-f][1,3,2]dioxaphosphepin (CPA-7) instead of CPA-2.

Fig. 4: Mechanism studies on the migratory cross-electrophile coupling via radical 1,2-N-shift.
figure 4

A Control experiments. B 11B NMR studies on the cleavage of B-B bond. C Proposed mechanism for the migratory coupling.

Mechanistic investigations

Next, control experiments were conducted to elucidate the cascade 1,2-N-shift/cross-electrophile coupling (Fig. 4). First, the use of the radical scavenger, 2, 2, 6, 6-tetramethyl-1-piperidinyloxyl (TEMPO•) was investigated. As expected, the 1,2-N-shift process was significantly suppressed, and HRMS analysis confirmed the formation of the TEMPO–1 adduct (53), suggesting the involvement of a radical-mediated process in the reaction (eq 1). In addition, a standard radical clock (cyclopropane) was subjected to the optimal reaction conditions, and none of the expected β-arylation of the amino acid ester 55 was observed. Instead, the formation of product 54 via a radical cyclopropane ring-opening process was observed by HRMS and 1H NMR (eq 2). This clearly indicates that the alkyl radical formed via dehalogenation by Ni likely triggers a sequential migratory cross-coupling reaction74. Notably, crossover experiments revealed a distinct mechanistic preference: the aminyl radical re-addition to acrylate predominantly occurs through an in-cage pathway, yielding products 3 (28%) and 41 (23%). In contrast, the out-of-cage addition products 56 and 57 were obtained in significantly lower yields (<10%) (eq 3). Based on our investigation of the cascade migration/cross-coupling of β-bromo amino acid esters, we discovered that 1,2-N-shift was beneficial in the kinetically favored N-extrusion/1,4-addition sequence, whereas 1,3-N-shift using a γ-bromo amino acid ester as the precursor appeared to be less effective (eq 4). Radical 1,2-N-shift/Heck-type coupling cascades were also observed when using α-trifluoromethyl alkenes as the Michael acceptors, which also suggested that an electrophilic α-carboxyl radical via 1,2-N-shift should generate in this radical relay (eq 5). In addition, The EPR spectrum of the reaction in the presence of NiBr2•dme/L1/K2CO3/B2neo2/1 with the spin-trapping reagent 5,5-Dimethyl-1-pyrroline-N-oxide, under standard conditions, indicated that some alkyl radical species were produced along with the desired product 3, confirming that a radical-mediated process was involved in the 1,2-N-shift process (eq 6). Furthermore, we performed a control experiment using aziridine as the radical precursor under the standard reaction conditions (eq 7). The results revealed that no corresponding product derived from a ring-opening/cross-coupling sequence was detected. This observation suggests that the radical-mediated nitrogen migration in our system may proceed via a mechanism distinct from those proposed by Doyle et al.75. Subsequently, the 11B NMR spectra confirmed that the Ni complex mediated the activation and cleavage of the B-B bond in B2neo2, thereby facilitating the generation of active Ni-Bneo species involved in the migratory coupling catalytic cycle, which aligns with a previous report (Fig. 4B)68,70.

Based on the experimental results and previously reported mechanisms12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78, the proposed mechanism for this cascade reaction is shown in Fig. 4C (CCDC 2383157 (for 3), CCDC 2383157 (for 6) and CCDC 2383157 (for 23) contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service). First, B2neo2 is activated by the Ln-NiIIBr2 complex with the aid of a base, thereby generating both the Ni-B complex, Ln-NiII(Bneo)Br, and carboxylic or bromoboronate (X-Bneo)68,76. Next, Ln-NiII(Bneo)Br undergoes reductive extrusion of the Br-Bneo species to deliver the LnNi0 species77,78,79, thus realizing the two-electron reduction of the Ni(II) to Ni(0) species in the Ni catalytic cycle. By contrast, the Ln-NiIBr complex undergoes a SET oxidation process with serine-derived bromide (1), yielding the radical intermediate Int-I and a Ni(II) intermediate, which then participates in the next catalytic cycle. Notably, the in situ-generated X-Bneo species could act as a Lewis acid to coordinate with the NR1R2 moiety in intermediate ‌int-I‌, which would contribute to the cleavage of the C(sp3)-N bond80. This mechanistic rationale provides a plausible explanation for the observed divergent migration preferences: while employing B2(neo)2 as the reductant favors radical 1,2-N-shift (as discussed in Table 1), the use of zinc powder selectively enables 1,4-aryl migration66. Subsequently, the NR1R2 species of the nascent intermediate int-I is extruded to form methyl acrylate (int-II) (In the reaction, GC-MS analysis successfully detected the formation of side-product olefin and (2,6-Me2C6H3CH2)NH during the N-extrusion process from the intermediate Int-1 to Int-2 in Fig. 4C, please see the ‘Control experiments’ section in Supplementary Information for details). This is followed by the 1,4-addition of the nascent •NR1R2 radical (likely exhibiting nucleophilic character81) to the resulting methyl acrylate to yield intermediate Int-III, thus realizing the formal radical 1,2-amino migration (see the ‘DFT Calculations’ section in Supplementary information for details on this “N-extrusion/1,4-addition process). During this DFT calculations, we attempted to locate an aziridine-type radical intermediate by cyclizing on the lone electron pair of the N-atom of Int-I, but were unsuccessful. One likely explanation for this failure is the high ring strain associated with the three-membered ring. Thus, our DFT calculations revealed that the cyclization on the lone electron pair of the N-atom of Int-I to produce an aziridine-type radical intermediate, enabling 1,2-N-shift could not occur. Notably, trans addition is favored during the re-addition process, which is consistent with the observation that this migratory coupling typically exhibits high diastereoselectivity (syn/trans > 20:1). In the cross-coupling catalytic cycle, the Ni0 species first undergoes oxidative addition to the aryl bromide to produce an ArNiII(Ln)Br complex. Subsequently, the α-ester radical (Int-III) generated by the 1,2-amino migration adds to ArNiII(Ln)Br to form a NiIII complex (Int-IV). Finally, the NiIII complex undergoes reductive elimination to furnish the desired product (3) and the Ln-NiI complex, which is directly involved in the next catalytic cycle. Notably, in the presence of water and a base, B2neo2 can undergo hydrolysis and liberate traces of hydrogen (see the ‘Control experiments’ section in Supplementary information for details)82, which appears to reduce NiBr2•dme to the Ni(I) or Ni(0) species to restart the Ni catalytic cycle.

Subsequently, our calculations also provide insight into the origins of stereocontrol in this migratory cross-coupling (Fig. 5). By using density functional theory (DFT) calculations (the computational methods are provided in the ‘DFT Calculations’ section in Supplementary information)83,84,85, we investigated in detail the origin of enantioselectivity in the radical addition step. Notably, the calculations were operated by using the chiral ligand (S,S)-m-tolyl-Box L7, methyl p-bromobenzoate, and serine derived bromide 1 as model substrate. Given that the Ni catalyst exhibits nearly identical free energies between its triplet and singlet states (with the triplet state being slightly more stable by 1.2 kcal/mol), we examined the radical addition to both states of the Ni catalyst. Transition states for the radical addition leading to both R- and S-intermediates were optimized. For each enantioselectivity, three distinct initial conformations of the interaction between the catalyst and the radical substrate were considered, resulting in a total of twelve optimized transition states (Table 2). As shown in Table 2, the transition states of S-TS2III and R-TS2III exhibit the lowest free energy barriers for the formation of S- and R-intermediates, respectively, with the latter being 0.6 kcal/mol more favourable. According to the Boltzmann distribution, this barrier difference corresponds to an approximate enantiomeric ratio of 3:1 favoring the formation of the R-intermediate, a prediction that aligns closely with experimental observations. Moreover, structural and non-covalent interaction analysis reveals that R-TS2III benefits from the reduced steric hindrance around the ester group of the radical substrate, resulting in a parallel arrangement and stronger dispersion interactions between the ester-radical conjugated system of the radical substrate and the Ni center of the catalyst. These factors are likely responsible for the observed enantioselectivity favoring the R-intermediate. Enlightened by Guo’s study on the synergistic catalysis—combining a phosphine−nickel complex and an axially chiral sodium dicarboxylate56, we conducted in-situ ³¹P NMR studies to identify chiral phosphoric acid (CPA-2) species participating in the asymmetric catalytic cycle (Fig. 6B). Initial characterization of CPA in CD3CN revealed two single resonances at δ = 6.94 ppm and 5.52 ppm. Subsequent neutralization with Na₂CO3 induced complete disappearance of the CPA signal, concomitant with the emergence of a new doublet (δ = 8.35 and 8.54 ppm) corresponding to phosphate formation. Remarkably, introducing migratory precursor 1a to the CPA/Na₂CO3 system maintained this doublet pattern (δ = 8.36 ppm and 8.54 ppm), suggesting no direct interaction between precursor 1a and the phosphate intermediate. Intriguingly, further addition of Ni(COD)2, L10, and methyl 4-iodobenzoate (in situ generating the Ar-NiL-X complex as shown in Fig. 6B) to the binary system of CPA and Na₂CO3 resulted in the emergence of a distinct ³¹P NMR resonance at ‌δ = 3.06 ppm‌. This shift, along with other spectral changes, indicates the formation of new nickel-phosphate complexes with structural modifications including ligand coordination mode, steric reorganization‌, etc. These observations collectively demonstrate that (1) CPA undergoes base-mediated conversion to phosphate species under catalytic conditions, and (2) synergistic coordination between the CPA-derived phosphate and the Ni/L10 complex generates a sterically defined chiral pocket. To further elucidate the enantioselectivity enhancement induced by CPA, we conducted additional DFT calculations on the CPA-mediated radical addition step. Analysis of the CPA-containing catalytic system showed that transition states R-TS2III.P and S-TS2III.P have the lowest free energy barriers for forming the S- and R-intermediates, respectively. R-TS2III.P is 1.8 kcal/mol more stable than S-TS2III.P (see ‘DFT Calculations’ section in Supplementary information for details). According to the Boltzmann distribution, this barrier difference corresponds to an approximate enantiomeric ratio of ~18:1 (R: S), a prediction that aligns with experimental observations. These ³¹P NMR and DFT results demonstrate that the synergistic effects between CPA phosphate and nickel/Box catalysis enable enantioselective control over the radical cascade, which involves a 1,2-N-shift and C(sp³)-C(sp²) cross-electrophile coupling.

Fig. 5: DFT optimized structures of the transition states.
figure 5

A Optimized structures of the transition states for the radical addition with the single-state NiIII complex as the active species. B Triplet-state NiIII complex as the active species.

Table 2 Properties of diverse transition states for the radical addition
Fig. 6: Non-covalent interaction (NCI) analysis and 31P NMR studies.
figure 6

A DFT investigation on the origin of enantioselectivity in the radical addition step: a Optimized structures of the lowest-barrier transition states for the formation of S-intermediates (S-TS2III) and R-intermediates (R-TS2III). b Their non-covalent interaction (NCI) analysis. In the structures, the distances are in angstrom (Å) and the unpaired spin populations are shown with the indication of “S”. In the NCI analysis, the green clouds (visualized as color-filled reduced density gradient isosurface) indicate dispersion interaction. B 31P NMR studies on the role of CPA in the asymmetric catalytic systems.

Discussion

In summary, we have disclosed a nickel/CPA co-catalyzed enantio- and regioselective migratory cross-electrophile coupling by π-system-independent 1,2-N-shift, allowing the straightforward synthesis of the non-natural α-arylated β-amino acid motifs through skeletal editing of β-bromo α-amino acid esters with aryl bromides in excellent high regio- and enantioselectivity. Upon cooperative catalysis of a nickel/chiral cyclo-Box complex and CPA, the radical cascades further realized enantioselective building of a C(sp2)-C(sp3) bond, thus accomplishing skeletal editing of racemic β-bromo α-amino acid esters toward enantioenriched β-amino acids. Notably, this success expands these strategies toward the challenging enantioselective migratory C(sp2)–C(sp3) cross-coupling at the alpha position of the ester moiety. In addition, the origin of the enantioselectivity in the radical addition step was clarified by DFT calculations. The radical relay is catalyzed by an asymmetric combination of a chiral cyclo-Box ligand and CPA, using organic diboron as a safe reductant and exhibits a broad scope and excellent enantio- and regioselectivity. The reaction is a powerful platform for the asymmetric synthesis of β-amino acid motifs from abundant and easily accessible α-amino acids.

Methods

Typical experimental procedure for migratory cross-coupling of β-bromo α-amino acid esters with aryl bromides

Method A

To a 10 mL Schlenk tube with screw-cap was added β-bromo α-amino acid ester (0.2 mmol, if solid), aryl bromide/tosylate (2 equiv, if solid), NiBr2(dme) (10 mol%), (12 mol%), L1 (12 mol%), K2CO3 (2 equiv), B2neo2 (2 equiv). The tube was evacuated and back-filled with argon  (this process was repeated three times), CH3CN/EtOH (v:v = 1.8:0.2) were added consecutively via syringe with β-bromo α-amino acid ester (0.2 mmol, if liquid) and aryl bromide/toyslate (0.4 mmol, if liquid). The resulting mixture was bubbled with argon  to degas for 1 min and stirred at 50 °C for 30 h. Note: a stirring speed above 800rpm is highly important for reproducibility. The resulted mixture was filtered through a short plug of silica gel to remove metal salts, and diluted with Et2O (50 mL). The filtrate was poured into a separatory funnel and partitioned with brine (20 mL). The aqueous layer was then extracted with diethyl ether (3 × 20 mL). The organic layer was dried over anhydrous Na2SO4 and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (eluent: n-hexane/ethyl acetate) to afford the corresponding products (3–46) in a yield as listed in Fig. 2.

Method B (Asymmetric)

To 3 mL vial were added sequentially NiBr2(dme) (10 mol%), (S,S)-L10* (12 mol%), (R)-CPA*-2 (10 mol%), β-bromo α-amino acid ester (0.1 mmol), aryl bromide (2 equiv), Na2CO3 (3 equiv), B2neo2 (2 equiv), and CH3CN (0.5 mL) in a dry glovebox. The vial was closed with a screw cap containing a PTFE septum, taken outside the dry box, and the solution of CH3CN (0.5 mL) with H2O (2.0 equiv.) were added consecutively via syringe under a positive flow of argon. The resulting mixture was stirred at 40 °C for 36 h. Note: a stirring speed above 800rpm is highly important for reproducibility. The resulted mixture was filtered through a short plug of silica gel to remove metal salts, and diluted with Et2O (30 mL). The filtrate was poured into a separatory funnel and partitioned with brine (10 mL). The aqueous layer was then extracted with diethyl ether (3 × 20 mL). The organic layer was dried over anhydrous Na2SO4 and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (eluent: n-hexane/ethyl acetate) to afford the corresponding enantioenriched β-amino acid esters in a yield as listed in Fig. 3.